Le 22/06/2015 14:14, Agustin Lobo a écrit :
I have a muti-spectral image with inter-band misallignmet and use i.e.
$ otbcli_HomologousPointsExtraction -in1 TTC4510v2.tif -band1 3 -in2
TTC4510v2.tif -band2 2 -mode full -precision 20 -mfilter 1 -algorithm
sift -out TTC4510v2homSIFT3.txt
to find homologous points.
The image is geotif with arbitrary coordiantes:
$ gdalinfo TTC4511v2.tif
Driver: GTiff/GeoTIFF
Files: TTC4511v2.tif
TTC4511v2.tif.aux.xml
Size is 1280, 1024
Coordinate System is `'
Origin = (0.000000000000000,1024.000000000000000)
Pixel Size = (1.000000000000000,-1.000000000000000)
Image Structure Metadata:
COMPRESSION=LZW
INTERLEAVE=PIXEL
Corner Coordinates:
Upper Left ( 0.000, 1024.000)
Lower Left ( 0.0000000, 0.0000000)
Upper Right ( 1280.000, 1024.000)
Lower Right ( 1280.000, 0.000)
Center ( 640.000, 512.000)
Band 1 Block=1280x1 Type=UInt16, ColorInterp=Gray
I'd like to confirm:
1. The resulting points in
-out TTC4510v2homSIFT3.txt
are x,y coordinates with (0,0) at the bottom left.
No, they will take into account origin and pixel size (i.e. physical space).
2. The argument
-precision 20
means that homologous points in image B (band 2 in this case) will be
searched within a radius of 20 pixels of points in image A (band 3 in
this case)
Not exactly. First, origin and pixel size of both images are taken into
account. Then :
- In geobins mode the precision gives the extra margin applied to bins
(i.e. roughly corresponding patches between images), to account for
metadata or sensor modelling precision,
- In both modes, if mfilter is on, matches that do not respect the
precision will be filtered afterward
3. The argument
-mfilter 1
means that couples of homologous points farer away than 20 pixels from
each other will be eliminated.
Yes, but this is in physical space.
When I display the points in QGIS I find them kind of weird. First, I
do not see them
as salient sites at all.
Do the match look correct ? This is SIFT/SURF magic : they are not
really related to landscape features.
Second, I do not see that they track the
displacement (there is
a rotation + horizontal and vertical displacement). This is why I'd
like to make sure I'm making a correct interpretation of the manual.
Well, if you have rotation and vertical displacement, that is not
represented by metadata or sensor modelling, you should not be using
mfilter option (unless you are sure that all matches should fall within
20 pixels of each other, whatever the distance to the rotation center).
--
Julien MICHEL
CNES - DCT/SI/AP - BPI 1219
18, avenue Edouard Belin
31401 Toulouse Cedex 09 - France
Tel: +33 561 282 894 - Fax: +33 561 283 109
--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html
You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
You received this message because you are subscribed to the Google Groups "otb-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
